| Projects |
| |
 |
Effective Shape Depiction
This is my dissertation research. Diverging from traditional approaches in computer graphics which use purely geometric
simplices as processing units, we argue that shape depiction can benefit from a
computation representation compatible with the human visual processing. To support
this argument, we first present a system that parses a 2D planar shape into a part-based
structure that approximately respects the structural organization in human perception.
And we show that simplifications on this representation align with the common
artistic practices in shape abstraction, in which only prominent parts are preserved and
the visual structures are more clarified than traditional simplification methods based
on local adaptation of geometric details.
|
|
| |
 |
Shape Understanding
Differing from traditional shape segmentation approach, we propose a new way to think how parts fit together by introducing a computational model that locates and describes the extend and the prominence of the transitional regions that parts connect to the rest of the shape. This idea of transition also gives rise to a unified mathematical characterization of local shape organization. In particular, it distinguishes extended wholes uniformly from various assemblies of parts, such as limbs, necks, and short-cuts.
|
|
| |
 |
Perceptually Guided Segmentation
This project studys how perceptual brightness and photometric measures could benefit the image segmentation. Instead of using these perception measures as pixel weights for segmentation, the proposed approach simply adds the perception measure as an additional dimension in the feature space. Taking human perception information into account makes the segmentation results better align to the human visual perception. Experiment results show that this approach produces aesthetic pleasing segmentation results, which is useful for further image based depiction.
- This is one of my computer vision course projects that I did mainly for fun. Feel free to download the source code and project report. Note that you need your own glut/glui library (I used the gluit maintained by Princeton folks included in Szymon's trimesh2)
|
|
| |
 |
Computer-Generated Papercutting
From the point of view of computer graphics, papercutting can be seen as a method of composing bi-level images under a set of geometric connectivity constraints. We present a technique for composing digital paper-cut designs. The elements of a design may be images, which are processed via a multi-layer thresholding operation, or they may be procedurally-generated arrangements of shapes. Elements are composed using a set of boolean operators that preserve connectivity. The resulting designs are well suited to being cut by a digital craft cutter to create real paper-cuts.
|
|
| |
 |
Calligraphy Simulation and Modeling
We propose a virtual brush model based on droplet operation for Chinese calligraphy and painting simulation. We use faster, hard accelerated droplet visualization for interactive virtual calligraphy and an elaborated, more time-consuming droplet rendering for more vivid results. Physically based bristle simulation is avoided to achieve significantly better performance while still maintaining expressiveness. Since the geometries of the boundaries of the brush trajectories are well defined, this work naturally extends to calligraphic modeling with abstracted expressiveness of dry strokes.
- Xiaofeng Mi, Jie Xu, Min Tang, Jinxiang Dong, The Droplet Virtual Brush for Calligraphic Chinese Character Modeling, Proceedings of the 6th IEEE WACV, Orlando, Florida, USA, December, 2002
- Xiaofeng Mi, Min Tang, Jinxiang Dong, Droplet: A Virtual Brush Model to Simulate Chinese Calligraphy and Painting. Journal of Computer Science and Technology, Vol. 19, No.3, Science Press, Beijing, China, Allerton Press, New York, USA , May 2004
|
|
| |
 |
User-defined Feature Modeling
This project proposes a distributed feature-based modeling service framework. An
interface model of feature-based modeling service towards such a modeling environment, including
the support of remote feature attachment and semantic maintenance is introduced. We use a user-friendly prototype-based feature
interface definition strategy. Mechanisms for procedural attachment, declarative validation, as well as some of the issues in distributed environment,
such as locality-transparent reusing, are
also explained in details.
|
|
| |
 |
Other Projects
This is an implementation of Bruce Gooch et al's Human facial illustration paper written in Java. You have to install the Java Media APIs to get the applet running. Another little program in Java is to dig useful information from ebay auction pages. email me for detailed instructions on how to get it compiling and running.
I have implemented quite a few research papers in computer graphics --- as a rest from getting stuck in my dissertation research:) This is an implementation of Eitan Grinspun's thin shell simulation. Instead of doing automatic derivatives, I derived everything by hand and here is the note. Takeo Igarashi's cool as rigid as possible shape manipulation also gets an implementation here. What's provided here merely helps you fill in the matrix, you will need to fill in your own class of point2D, matrix solvers, etc to get it to work, which should be very easy. There is an implementation of GPU accelerated suggestive contour rendering here. Enclosed are the vertex and fragment shader files, you will need to send in the principle curvatures and directions, as well as derivatives of curvatures to the shaders. Szymon's trimesh2 library computes everything. The nice thing about my shader is that it has a very good control of screen space line width - which is not trivial at all to get right. Email me if you are interested in the details in how differential geometry plays a role here. Also, here is some texture synthesis I implemented long time ago. |
|
|
